Journal of International Economics 70 (2006) 406 – 427
www.elsevier.com/locate/econbase
Globalization and the ‘confidence game’
Sharun W. Mukand ⁎
Tufts University, Department of Economics, 305 Braker Hall, Medford, MA 02155, United States
Received 30 October 2000; received in revised form 16 September 2005; accepted 1 March 2006
Abstract
We show that governments in developing countries have an incentive to play the “confidence game” —
wherein the need to win the confidence of the international capital market ‘can actually prevent a country
from following otherwise sensible policies and force it to follow policies that it would normally consider
perverse’. This incentive arises because of a combination of a ‘conformity bias’ and ‘good news bias’ in
governmental decision making in an open economy, which results in inefficient outcomes which increases
rather than decreases the threat of devaluation. While institutions that encourage greater transparency and
the public revelation of information, may often mitigate this inefficiency, on some occasions increased
transparency may even exacerbate the inefficiency.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Globalization; Devaluation; International capital markets; Currency crisis; Transparency
JEL classification: F33; F40
1. Introduction
The globalization of capital flows in recent years, has increased access to international capital
markets for many developing countries. This globalization is believed to have the potential to
discipline governments in developing countries. This is because it is perceived that in order to
attract foreign capital, governments have to pursue ‘sound’ economic policies in order to boost
‘market confidence’ (Camdessus, 1998).1 While this view seems quite plausible, the recent spate
⁎ Fax: +1 617 6273137.
E-mail address: sharun.mukand@tufts.edu.
1
This view has been echoed by Obstfeld (1998), Feldstein (2000) and Summers (2000). For another instance consider
Dornbusch (2000) who says that on facing a crisis, a finance minister will “have the ambulance rush them to the IMF. And
when they do, markets start taking confidence very soon and from there it is a short step to normalization” (emphasis
added). An excellent discussion of the issues involved in Krugman (1999) and Rodrik (1999).
0022-1996/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jinteco.2006.03.001
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
407
of currency crises seems to suggest that the ‘disciplining’ role of the international capital market
has at the very least, not been very effective.
In a perceptive commentary Krugman (1998, 1999) suggests a potential reason. He argues
that inefficiencies might arise because policymakers in the global economy feel compelled to
play the ‘confidence game’ — where the preoccupation of policymakers is not with economic
fundamentals but with winning market confidence. There is a temptation to implement “…
policies that may not make sense in and of themselves but that policymakers believe will appeal
to the prejudices of investors…In fact the need to win that confidence can actually prevent a
country from following otherwise sensible policies and force it to follow policies that would
normally seem perverse.”2 This opinion has been echoed by Bhagwati (1998), Rodrik (1998)
and Stiglitz (2002) who also conjecture that globalization and the free flow of capital has
resulted in developing countries feeling ‘compelled’ to enact policies that may be inappropriate.
This is somewhat easier to understand if there is a fundamental dissonance between the
developmental goals of a country and the interests of foreign investors.3 It is more puzzling
once we recognize that both — a productive economic environment and increased foreign
capital, raise national income. However, in this paper we argue that once we take account of
the signaling aspect of policy choice, the desire to attract foreign capital rather than ‘discipline’
governments, generates perverse incentives for policymakers and results in inefficient
outcomes — making currency crises more rather than less likely.4 In that sense this paper
provides an analytical underpinning to the opinions voiced by Krugman, Bhagwati and Rodrik
in recent years, that globalization and the free flow of capital might generate perverse
incentives for governments in developing countries.
In our framework, a government that is interested in defending a fixed exchange rate, chooses
between policies. Two features of our framework are worth highlighting. First, policies are
assumed to be state-dependent. We assume that different policies are ‘appropriate’ for different
states of the world — no policy is intrinsically superior to the other. Therefore, if the government
has a very strong foreign reserve situation, the ‘appropriate’ policies are likely to be very different
from the policies that it should pursue when its foreign reserve situation is precarious. Given this
state-dependence, policymaking for the government would seem to be a simple matter of
matching the ‘appropriate’ policy to the corresponding underlying state. However, what
complicates matters is that the government does not know for sure, the underlying state of the
world. This brings us to the second important aspect of our framework. In particular, the
government has reliable (even if not perfect) private information about the underlying state of the
world. In this framework, we show that with the globalization of capital flows, governmental
decision making in developing countries might suffer from inefficiencies. In particular, we argue
that, the very attempt to maintain ‘market confidence’, may result in governments feeling
‘compelled’ to enact (or persist with) policies that have an adverse impact on the economic
environment.
We identify two situations under which this might happen and governments have an incentive
to play the ‘confidence game’. The first reason why governments play the ‘confidence game’
2
This quote is from Krugman (1998).
Rodrik (1998) argues that the free flow of capital can make policy making in “..‘emerging market’ countries hostage
to the whims and fancies of two dozen or so thirty-something country analysts in London, Frankfurt and New York. A
finance minister whose top priority is to keep foreign investors happy will be one who pays less attention to
developmental goals”.
4
Our analysis also helps throw light on the alleged ‘failure’ of economic policies in stemming the South East Asian
crisis. See Radelet and Sachs (1998) and Wade and Veneroso (1998) for an elaboration.
3
408
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
arises when priors are skewed in favor of a particular state of the world. So if the priors of the
international capital market are that a tight fiscal policy is the correct policy, the government will
be ‘compelled’ to slash budget deficits, even if it had private information that running a temporary
budget deficit might be more suitable. By enacting a policy that conforms with the priors of the
international capital market, the government signals that it is confident of success and hopes to
attract foreign capital. If the government through its policy choice, manages to attract sufficient
amount of foreign capital, so as to more than offset the adverse impact of a poorer economic
environment. We call this the ‘conformity bias’ in governmental decision making. In a sense this
conformity bias captures Krugman's (1998) worry that in the new global economy “sound
economic policy is not sufficient to gain market confidence; one must cater to the perceptions, the
prejudices, and the whims of the market.” It is quite striking to observe that, the government's
attempt to enhance its ability to defend the exchange rate by playing the confidence game, makes
it more rather than less vulnerable to a currency crisis. We further show that investment flows
have a nonmonotonic relationship with the degree of ex–ante uncertainty about the possibility of
success of any given policy. Investment increases as uncertainty decreases until a certain
threshold level of uncertainty is reached. A further decrease in uncertainty results in inefficiency
and a collapse of capital flows because it results in a ‘conformity bias’ in governmental decision
making.5
We then demonstrate another scenario in which a government may mislead foreign investors.
Consider the case where priors are not skewed, but that the productivity of the economic
environment differs across the two states. In this scenario we show that the government will have
an incentive to enact a policy that signals that ‘good news’ lies ahead i.e. productivity is going to
be high since good times are expected. By acting confident and reflecting its inherent optimism
about the underlying fundamentals in its policy choice, the government hopes to convince foreign
investors to invest more than what they would have otherwise. We call this the ‘good news bias’ in
government decision making. The government will have a tendency to hope for the best and
pretend that business is as usual, rather than make the necessary policy adjustments to minimize
the risk of a devaluation.
Do mechanisms that may mitigate the government's incentive to deceive investor and play
the confidence game exist? We explore this question by examining the impact of greater
transparency on government policy choices. At one level the answer to this question seems
straightforward — as argued by Rogoff (1999) ‘increased transparency would undoubtedly be
useful in achieving more efficient global markets’. Indeed institutions such as the IMF share this
belief in the efficacy of greater transparency in ensuring efficient government policy choices.
For instance, in response to the Mexican currency crisis, the IMF established mechanisms for
greater transparency such as the SDSS and GDSS (see Fischer, 2002). Our analysis suggests
that while greater transparency typically does improve matters, there may be some exceptions.
Indeed in some cases, it is precisely the increase in public information availability that result in
governments having an incentive to play the confidence game — worsening matters. This
somewhat striking possibility is in accord with recent work on the social value of public
information (Morris and Shin, 2002).
This paper adds to the nascent literature examining various aspects of globalization. Our
analysis suggests that if globalization of capital flows proceeds far enough, then governments
may have an incentive to play the confidence game — lowering capital inflows and weakening
5
It is important to remind ourselves that all the results that we highlight take place only when globalization has
proceeded far enough to allow for the free flow of capital.
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
409
fundamentals. Therefore, with globalization, some countries become more rather than less
vulnerable to the threat of devaluation. Our analysis complements other recent work on some of
the potential adverse effects of alternative aspects of globalization. For instance, Krugman and
Venables (1995) examine the role of a reduction in transport costs on welfare. Cai and Treisman
(2005) suggest that heterogeneity in institutional structure across countries, then greater capital
mobility may weaken discipline. Mukand and Rodrik (2005) show how inefficiencies may arise
when we account for the impact that global informational spillovers on policymaking of
neighboring countries.
Furthermore, our work is related to the literature on policy signaling in an open economy.6 For
example, Bartolini and Drazen (1997) study the impact of imposition of capital flows by focusing
on the signaling aspect of such a policy choice. The government's decision of whether to impose
capital controls is observed by investors, who use it to infer about the future course of such policies.
However, in our paper the observed policy choice has no dynamic implications about the future
course of policies, since the emphasis over here is on the signaling of private information, and the
perverse incentive effects this generates. Finally, this paper is also similar in structure to work by
Prendergast and Stole (1996), Majumdar and Mukand (2004), and especially Brandenburger and
Polak (1996). Under the assumption that managers care about the current valuation of firms, in the
latter paper, there emerges an informational inefficiency similar to the one in this paper. There are
several important differences, however. In our model the focus is on the interaction between
governmental policy, exchange rates and international capital flows. In sharp contrast to their
paper, our results imply that the informational inefficiency may persist even if managers care about
the value of the final output, the assumption of current valuation is not essential.
In Section 2.1, we set up the benchmark economy and analyze the equilibrium. Section 2.2
elaborates on the basic model and describes a specific application to fiscal policy. Section 3
explores the potential impact of international institutions in increasing transparency. Section 4
concludes with a discussion.
2. The framework
We present the basic framework in two stages. In Section 2.1 we present a stripped down
version of the basic model, that has that advantage of presenting our basic results in a transparent
way. In Section 2.2, we further explore the model by placing it in the context of government
expenditures and deficits.
2.1. The benchmark model
We adapt to a signaling context, the exchange rate framework developed by Morris and Shin
(1998). The model highlights the strategic interaction between foreign investors and a
government interested in defending an exchange rate peg. The government's ability to defend
the exchange rate is a function of the strength of the country's underlying economic
fundamentals. Through its policy choice the government has the potential to weaken or
strengthen the country's “fundamentals” — thereby either lowering or exacerbating the threat of a
currency collapse. The difficulty with policymaking is that there is uncertainty about what
constitutes the appropriate policy. An inappropriate policy choice by weakening fundamentals,
may increase the threat of an exchange rate collapse. However in the first period, the government
6
See Drazen (1997) for an excellent survey of the nascent literature on this topic.
410
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
Nature chooses
state of world S j
with prior
probability ω.
Govt. receives
private signal and
updates prior about
state of world. It then
chooses policy x j
Foreign investors
observe xj and
choose investment
levels into country
Public good
productivity, output
and the random var. z
are realized. Govt.
makes decision of
whether or not to devalue.
Payoffs
Realized
Fig. 1. Timing of the game.
receives a private signal about the appropriate policy choice. The government updates its prior
about the appropriateness of alternative policy choices and then chooses a policy. Foreign
investors observe the government's policy choice and choose an investment level. Depending on
the amount of foreign capital it attracts, as well as the direct impact of the policy on the economic
environment, the government decides whether to defend the exchange rate or not (see Fig. 1 for a
schematic).
We now describe the features of the model sketched out above in greater detail.
Exchange rates and ‘fundamentals’: The economy in period t is characterized by a state of
economic fundamentals θt. A country's economic fundamentals in any period are not exogenous,
but rather are influenced by government policy choices, foreign investment, exogenous economic
shocks as well as the state of fundamentals in the previous period. Accordingly, we assume that
the state of fundamentals in period t, are a function of both the fundamentals inherited from the
previous period as well as current national output yt such that θt = θ(θt−1,yt), where θt is increasing
in both the arguments.7 Therefore, a higher value of θ corresponds to ‘stronger’ fundamentals. We
interpret these economic fundamentals as a proxy for any or all factors that affect a country's
ability to defend its exchange rate. For instance, they may be interpreted as being equal to the
level of foreign exchange reserves of the government (as in Section 2.2 of the paper), or a
country's national income or its level of foreign debt or even the country's overall
macroeconomic situation.
An economy's fundamentals matter because they influence the exchange rate. In particular, in
the absence of government intervention, a country's exchange rate is an increasing function of its
economic fundamentals θ and is given by f(θ). Accordingly, we assume that the government
pegged the exchange rate in period t − 1 at e⁎, where e⁎ = 1 ≥ f(θt−1) and the exchange rate is
defined in terms of the ratio of foreign to domestic currency. The government's ability to defend
its exchange rate peg depends on its policy choices.
Government policymaking in an uncertain world: There are two important aspects of
policymaking that we want to highlight. First, is the uncertainty inherent in much of
policymaking. Second, we are also interested in capturing the state-contingent nature of many
policies where different policies are appropriate in different contexts.
We capture both these elements by assuming that while different policies are appropriate
for different underlying states of the world, this state of the world is imperfectly known. In the
face of this uncertainty, the government chooses a policy xj from a binary set of policies each
of which is an ‘appropriate’ response to the underlying state of the world Sj, where j ∈ {L,H}.
Depending on the economic context, this policy can be variously interpreted as a conservative
or liberal budget policy, a high or low interest rate or more generally any and all policies
7
As will soon become clear, this formulation helps capture Obstfeld's (1994) notion that negative shocks can (by
weakening an economy's fundamentals) — help trigger devaluations.
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
411
which may have a stochastic impact on the economy.8 For expositional purposes we
deliberately keep both the nature and the impact of these policies somewhat abstract — further
elaborating in a concrete macroeconomic context only in Section 2.2. Accordingly, we assume
that if policy xj is correctly matched with the appropriate underlying state of the world — then
it has a positive impact on the economic environment since it results in public good
productivity of Gj. In contrast, if policy xj is ‘inappropriately’ matched to the underlying state
wk, then this adversely affects public good productivity, which now equals G0, where
GH ≥ GL > G0 ≥ 0 and j ≠ k.
Policy success and the role of information: The government's policy decision is made
difficult by the fact that when it chooses its policy, even though it has a prior ω(= P(SH)) that
the state of the world is SH, it does not perfectly know the true underlying state of the world.
However, the government receives a private signal sj, that the unknown state is Sj, where
j ∈ {L,H}. The degree of reliability or the likelihood that the private signal is an accurate
reflection of the true shock is ϕ, where ϕ = P(sH|SH) = P(sL|SL). Given this private signal, the
government uses Bayes rule to update its priors on the true state. Therefore, if the private
signal received is sH and its degree of reliability is ϕ we have the government's posterior
/x
on the state, which is, PðSH jsH Þ ¼ /x þ ð1−/Þð1−xÞ
. Similarly, if the government receives the signal sL
/ð1−xÞ
we have, PðSL jsL Þ ¼ /ð1−xÞ þ ð1−/Þx.
Now consider the relationship between the private signal and the prior. There exist two
possibilities, either the private signal is more reliable than the prior or vice versa. We focus on the
case where the degree of reliability of the private signal received by the government ϕ, is greater
than that of the prior ω.9 This case is of interest since it illustrates the potential for conflict
between the government and the foreign investors. We impose parametric restrictions such that
this is in fact the case.
Assumption I. The degree of reliability of the private signal is always greater than the prior, i.e.
1 > ϕ > ω ≥ 1 / 2.
Simple manipulation of the above updated probabilities enables us to state the following
Lemma,
Lemma I. If 1 > ϕ > ω > 1 / 2 then P(SH|sH) > P(SL|sL) > 1 / 2.
Therefore conditional on receiving the private signal si, the probability that the unknown
state is Si, is always higher when the signal is sH than when it is sL. Notice that posteriors are
‘unbalanced’ because the government's private signal is stochastic and not perfectly reliable
(i.e. ϕ < 1). Assumption I in conjunction with Lemma I gives rise to the following observation.
Observation I. The government's choice of policy xi when the private signal is si, i ∈ {L,H}, will
maximize expected productivity of the economic environment (public good).
8
So for instance in some states of the world, running a budget deficit may well be the ‘appropriate’ policy, while in
other contexts a fiscally conservative ‘balanced budget’ may be the correct policy. Similarly, introducing a currency board
may the correct policy for a country under some conditions while dollarization may have been appropriate in other
contexts. Mukand and Rodrik (2005) give numerous examples of the state and context dependent ‘specificity’ of policies
and their analysis helps generate a plausible accounting of the economic performance across transition economies.
9
The other possible case is that the degree of reliability of the prior ω is greater than the private signal si received by
the government i.e. ω > ϕ > 1 / 2. Therefore, quite irrespective of the private signal si received, the government's policy
choice will be to go by its prior if it wants to maximize the productivity of the public good and there is no potential source
of conflict between the government and foreign investors.
412
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
Therefore, if the government wants to choose a policy that maximizes the productivity of the
public good, it's choice of policy should always be in accordance with its private information.
Government and foreign investors: This choice of policy is important not just because of its
impact on the productivity of the economic environment, but also since it has an impact on the
government's ability to defend the exchange rate as well as national income. The government's
objective function is kept simple, and we assume that while the government derives a value V > 1
from successfully defending the exchange rate at the pegged level, it also incurs some costs, such
that the government's payoff is given by Wg = V / C. This formulation captures the notion that the
government obtains a lump sum payoff from not engineering a surprise devaluation — since it
avoids a loss of pride, voter disapproval and even removal from office.10 This cost of defending
the exchange rate is decreasing in the amount of foreign investment and the size of national
income and equals C = Z / [e(yt+θt− 1)], where yt is total output and Z > 1 is a random variable drawn
from a symmetric single peaked distribution.11 Therefore taking logs we obtain the government's
payoff from successful defense of the exchange rate is given by (similar to Morris and Shin, 1998)
wg = v − c = v + θt−1 + yt − z.12 Therefore, observe that the government's payoff from a successful defense of the exchange rate equals v − c, which is increasing in the size of the national income and foreign
investment. In contrast, the government's payoff from abandoning the fixed exchange rate is zero.
We assume that z is a random variable whose realization takes place just prior to the
government's decision to devalue or maintain the exchange rate. As such z is a proxy for all the
unexpected exogenous factors that may the government more or less be willing to defend the
exchange rate. As a result of this formulation, the probability of devaluation becomes a smooth
increasing function of capital inflows — we elaborate on this later. If the payoff that is realized by
the government is v + θ̂ t ≥ ẑ , then the government will successfully defend the exchange rate,
otherwise it will devalue and obtain a zero payoff.
Output is a function of the productivity of the public good and the total amount of capital
invested (k) in the country by a continuum of foreign investors, and is given by y = Gjg(k),
where g′ > 0, g″ < 0. The amount invested in the country is a simple decreasing function of the
opportunity cost of capital given by the world rate of interest r, which we denominate in terms
of the domestic currency.
The timing: The above benchmark model demonstrates the interaction between foreign
investors and the government which pegs the exchange rate in accordance with the prevailing
state of fundamentals (see Fig. 1). In the first period, the government receives an informative
private signal about the underlying state of the world. Accordingly, the government updates its
priors about the likely success of alternative policy choices and implements a publicly observable
policy xj, which affects the returns to foreign investment. Foreign investors observe the policy
enacted by the government and choose investment levels. Output is realized and so is the random
variable z. As a function of whether wg is positive or not, the government decides whether to
defend the exchange rate or to let it devalue. Finally, the game ends as the government and foreign
investors realize payoffs.
10
In making this assumption we follow a long tradition following Obstfeld (1994, 1996), Ozkan and Sutherland (1995),
Sachs et al. (1996) and Morris and Shin (1998) and Angeletos and Werning (2005). de Kock and Grilli (1993) show how
a similar formulation may emerge from private sector trigger strategies where the government is punished for deviating
from a fixed rate.
11
This particular formulation is for simplicity and nothing crucial hinges on it — all that is important is that the cost of
defending the exchange rate is decreasing in the strength of inherited fundamentals θt−1 and the total output.
12
Accordingly for the rest of the paper, we use lower case Roman numerals to denote natural logarithms.
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
413
2.1.1. Confidence and deception: equilibrium analysis
2.1.1.1. The public signal case. We begin by considering the case where the signal received by
the government sj, is publicly observed by foreign investors. The important thing to notice is that
the expected productivity of the public good (and hence returns to foreign investment) is higher
when the signal received is sH rather than sL.
In order to see this let us describe the foreign investor's optimization problem under the
assumption that he is atomistic and believes that the government's equilibrium policy strategy
σ⁎(xj,si) is to enact policy xj when it receives the signal si. Then we have,
max ½ð1−PðdjSj ; xj ; si ÞÞ þ PðdjSj ; xj ; si Þejt̂ PðSj jsj ÞGj gðkÞ þ ½ð1−PðdjSm ; xj ; si ÞÞ
k
þ PðdjSk ; xj ; si Þekt̂ G0 gðkÞ−ð1 þ rÞk:
Here P(d|Sm,xj,si)) is the probability of a devaluation of the currency to some êt < 1, if the
government chooses policy xj on receipt of signal si when the underlying state is Sm. Of course, if
there is no devaluation, then the exchange rate remains anchored at the prevailing exchange rate
of e⁎ = 1.
The investor's first order condition gives the optimal amount of foreign investment when the
government's optimal strategy is to pursue policy xj when it receives signal si is given by kji⁎(P(Sj|
si)Gj,r). Observe that in a world with public information, the amount of foreign investment is
increasing in the expected productivity of the public good. This suggests that in a world with
public information, there is no inefficiency in government policy choices.
Proposition I. If there is no private information, then for all ω ∈ (1 / 2,1] and Gi > Gj there exists
a unique equilibrium where the government selects policy xj when the signal is sj where i, j ∈ {H,
L}.
Proof. See Appendix.
□
Observe that the higher capital inflows generated when the government enacts policy xH rather
than xL, may arise due to either of the following reasons. First, the initial skew in the priors (i.e.
ω > 1 / 2), implies that we know from Lemma I that the posteriors are unbalanced and P(SH|sH) > P
(SL|sL). Second, the relative productivity of the public good may also be higher when policy xH is
enacted, since conditional on success, policy xH is intrinsically more productive than policy xL,
i.e. GH > GL. The key aspect to keep in mind is that when there is no private information, the
objectives of the government and foreign investors are aligned, and by ignoring the signal, the
government hurts not just the investors, but also itself — by exacerbating the chances of
devaluation. Therefore, through its policy choice, the government always maximizes public good
productivity. However, in the presence of private information, this alignment of the preferences of
the investor with those of the government may break down. We consider this possibility next.
2.1.1.2. Equilibrium with private information. Government policies have a direct impact on
the returns to investment, through their effect on the economic environment. However, as we
demonstrate below, the influence of government policies goes beyond this direct impact. Since
the government has private information about the underlying state of the world, its choice of
policy can convey a signal to foreign investors about the expected productivity of the public
good. Once we take cognizance of this signaling aspect of policy choice, we see that
governments may have an incentive to deceive foreign investors.
414
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
Recall our analysis of the economy with no private information. We saw that the expected
productivity of policies (and returns to foreign investors) are higher when the policy chosen was
⁎ > kLL
⁎ . This asymmetry in capital flows across the two policies,
xH rather than xL, resulting in kHH
is key to our argument, since it gives governments an incentive to deceive investors and enact
policies that do not maximize the returns to investment.
To see this clearly, examine the following scenario. Consider the government's payoff from
deviation from a proposed truth-telling equilibrium where its policy choice always pursues a
policy in accordance with its private signal (maximizes expected public good productivity) i.e.
government's equilibrium strategy is σ⁎(xj,sj). In our formulation, if the government successfully
manages to defend the exchange rate peg, when its inherited fundamentals equal θ̂ t−1, then its
payoff is given by wgt = v − c = v + θ̂ t−1 + yt − z. In contrast, the government's payoff if it fails to
defend the exchange rate, is zero. Now under the assumption that the government's equilibrium
strategy is to follow its private signal (i.e. σjj⁎), its payoff from pursuing policy xL after receipt of
sL is given by,
⁎ Þ þ w0 ½v− z̄ þ h ̂
̂ þ PðSL jsL ÞGLg ðkLL
wg ðxL ; sL ; r⁎jj Þ ¼ wLLL ½v− z̄ þ ht−1
t−1
LL
⁎ Þ:
þ ð1−PðSL jsL ÞÞG0 gðkLL
ð1Þ
Here for brevity, we define ψjim = [1 − P(d|xj,si,Gm)], which is the probability of no devaluation
occurring when as a function of its equilibrium strategy, the government chooses policy xj on
receiving si and realized productivity is Gm. The first term on the right hand side is the payoff if
the realization of public good productivity is GL, while the second term is the government's
payoff if policy is inappropriately matched with the underlying state, resulting in public good
productivity of G0.
In contrast, the government's incentive from deviating from the proposed truth-telling
equilibrium and choosing policy xH on receipt of sL is,
0
⁎
̂
̂
wg ðxH ; sL ; r⁎jj Þ ¼ wH
HL ½v−z þ ht−1 þ PðSH jsL ÞGH gðkHH Þ þ wLL ½v−z þ ht−1
⁎
þ PðSL jsL ÞG0 gðkHH Þ:
ð2Þ
The preceding two equations immediately imply that the government will always have an
incentive to deceive foreign investors and deviate from a strategy of always maximizing the
productivity of the economic environment, so long as wg(xH,sL;σjj⁎) ≥ wg(xL,sL;σjj⁎). On
simplifying (see Appendix), we observe that the government does not choose the first best
policy because the government fears that by doing so it may have an adverse impact on
international capital markets. Foreign investors will invest less in the country since they are less
‘confident’ that government policy will be successful and result in high returns. This loss in
‘market confidence’ can have two sources — first, the skew in the priors (i.e. ω closer to one), and
second, public good productivity across the two states. In both these cases, the government's
incentive to deceive arises since by enacting xH, the gain in foreign capital inflows more than
offsets the loss in income because of the lower (expected) productivity of the public good. We
now elaborate on both these reasons below.
2.1.2. The ‘conformity bias’ equilibrium
Consider first the possibility of a pooling equilibrium in which the government's incentive to
deliberately ignore its private information arises solely because of a skew in its prior. Accordingly
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
415
we assume that GH = GL and P(SH) > 1 / 2. In the following proposition, we demonstrate the
possibility of existence of such an equilibrium. Proof of this and subsequent propositions are
available in the Appendix.
Proposition IIA. Government behavior in pure strategies is characterized by the following,
(i) Conformity bias: ∃aω̄ such that ∀ω ≥ ω̄ , we have a pooling Perfect Bayesian equilibrium
where the government in its policy choice ‘conforms’ with the priors of the market and
enacts policy xH, when GH = GL.
(ii) ∃aω̂ such that ∀ω < ω̂ we have a separating equilibrium where the government enacts
policy xi when it receives the private signal si.
The intuition for the above proposition is best understood by considering the following
scenario. Consider the government's dilemma when it receives the private signal sL that
contradicts its priors about which policy is best. If the government undertakes the policy that
maximizes the returns to capital, from Observation I it will choose policy xL that is in
accordance with its private signal. However, since such a policy contradicts its prior, the
government (and the investors) are less ‘confident’ of its eventual success (i.e. posterior closer
to 1 / 2). Therefore, foreign investors are likely to invest a lower amount in the country. On the
other hand, if the government chooses a policy that ‘conforms’ with the priors of the
international capital market, it signals that it is relatively ‘confident’ of success. Therefore, by
‘conforming’ to the priors of the international capital market, the government hopes to signal
confidence about success, and deceives investors into investing a large amount of capital into
the country. Of course, such a policy may adversely affect fundamentals and even precipitate an
exchange rate devaluation. However, so long as the capital inflows are large enough, it is a
worthwhile gamble.
The above result is predicated on the assumption that globalization of capital flows has gone
far enough. This is because it is only with sufficient globalization, will the government's ability to
attract foreign capital be strong so as to give it an incentive to play the confidence game.
Unfortunately for the government, in equilibrium, its incentive to play the confidence game is
perfectly anticipated by foreign investors — depressing foreign capital inflows. This has a
striking implication: the government's attempt to attract capital to strengthen fundamentals and
increase its ability to defend the exchange rate, ends up increasing the likelihood of an exchange
rate collapse.
This proclivity of the government to play the confidence game, is seen best when we
examine the relationship between uncertainty and capital inflows. In what follows, it is useful to
keep in mind that in our analysis, equilibrium capital flows and welfare are positively related.
We begin by calculating expected capital flows. In a separating equilibrium where the
government selects a policy in accordance with its private signal we have to account for the
probability of signal si being received by the government, i.e. P(sH) = ωϕ + (1 − ω)(1 − ϕ). Then
⁎ (σjj⁎) + (1 − P
⁎ = P(sH)kHH
expected capital inflows in the case of a separating equilibrium Eksep
⁎
⁎
(sH))kLL (σjj ). Similarly, expected capital inflows in the case of a pooling equilibrium, where the
⁎ (σHj
⁎ ),
⁎ (xH) = kHH
government takes xH regardless of the private signal received is given by Ekpool
where capital flows are a function of the government's equilibrium strategy. We now use this
information to graph the relationship between expected capital inflows and changes in the degree
of uncertainty in Fig. 2.13 Observe that Fig. 1 demonstrates a nonmonotonic relationship with the
13
Fig. 2 is drawn for ϕ close to 1 and ω ∈ [0,ϕ) and GH = GL.
416
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
Expected
Capital
Inflows
Conformity
Bias exists:
Pooling
equilibrium
Govt. enacts
policy that
maximizes
returns:
Separating
equilibrium
~ ^ _
ωω ω
1/2
1
UNCERTAINTY
ABOUT STATE
OF WORLD - ω
(φ 'close' to 1)
Fig. 2. Uncertainty and the “conformity bias”.
degree of uncertainty about the underlying state of the world. When uncertainty about the state of
the world is low enough, the government's equilibrium strategy maximizes the returns to
investment. As this uncertainty comes down (i.e. ω decreases), the prior information available to
the government improves and the government's policy choice is more likely to be successfully
matched to the underlying state. This stimulates investment. Since over this range, the government
has no incentive to deceive, ex–ante capital inflows rise with a decrease in uncertainty.
However, if uncertainty decreases beyond a threshold ω̄ , we have a different scenario.
The government now switches its equilibrium strategy. It now has an incentive to ignore its
private information, and chooses policy xH quite irrespective of the private signal received.
Foreign investors are aware of this, and choose (in equilibrium) to go by their priors in
making investment decisions. The government's attempt to induce an increase in
investment, only succeeds in depressing it. Further decreases in uncertainty about the state of
the world, implies that even if the government always chooses policy xH, it is (in expected
terms) more likely to be successful. Therefore investment gradually rises. However, investment
remains lower than what it would have been if there was no conformity bias pooling
equilibrium.
2.1.3. The ‘good news’ bias equilibrium
Further, we identify a second set of parameters for which the government has an incentive to
play the ‘confidence game’. In particular, examine the case where international capital markets do
not expect one policy as more likely to be ‘appropriate’ than the other i.e. priors are balanced such
that P(SH) = P(SL = 1 / 2). We demonstrate that even in this case, so long as the relative productivity
of the public good differs across the states (i.e. GH > GL), we have a pooling equilibrium in which
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
417
the government ignores its private information. We call this the good news bias in governmental
decision making.
Proposition IIB. Government behavior in pure strategies is characterized by the following,
(i) Good news bias: ∃aḠ H such that ∀GH ≥ Ḡ H, we have a pooling Perfect Bayesian
equilibrium where the government in its policy choice enacts the policy xH, when
ω = 1 / 2.
(ii) ∃aG¯H such that ∀GH < Ḡ H we have a separating equilibrium where the government enacts
policy xi when it receives the private signal si, where GH ∈ [GL,Ḡ H) and ω = 1 / 2.
The intuition for the above result is simple. The government is aware that capital inflows
will be higher and it will be easier to defend the exchange rate if foreign investors believe that
the economy is likely to pick up and good times are ahead. Therefore through its choice of
policy xH, the government attempts to signal to foreign investors that it expects productivity of
the public good to be high (i.e. GH) rather than low.14 This results in a pooling equilibrium
driven by a ‘good news’ bias. Once again since in a ‘good news bias’ equilibrium, the
government does not utilize its private information, the capital inflows are depressed —
weakening fundamentals and increasing the likelihood of a currency collapse.
In the benchmark model that we have developed, we have not elaborated on the nature and
economic context of policies that may affect private returns. This had the merit of keeping the
analysis relatively transparent, and also facilitated analytical disentangling of the ‘conformity
bias’ from the ‘good news bias’ — in the process demonstrating how both these effects could
independently result in inefficient policy choices. This is useful since in many real world
applications, while both these effects may be present, they are difficult to distinguish and indeed
may also be working in opposite directions. This is brought home when we apply our framework
to throw light on aspects of some recent currency crises.
2.2. The extended framework: application to balanced budgets
In his commentary on the spate of currency crises that occurred in the nineties, Krugman
(1999) suggests that governments in developing countries interested in maintaining their
exchange rate have an incentive to play the confidence game. In particular he suggested that the
preoccupation with following ‘sound’ economic policies such as reduced government spending
and balanced budgets, may not only hurt fundamentals but may even precipitate currency crises.
In this section we sketch a simple application of the framework developed in the previous section
by describing a stripped down model of government fiscal policymaking.
A government's choice of an aggressive or conservative fiscal policy is a function of its foreign
exchange earnings which can be low RL or high RH, where RH > RL.15 If foreign exchange
earnings are expected to be high, then higher government expenditure strengthens the economy
14
Of course, if the productivity of GH rises and becomes much higher than GL, though the government will continue to
pool, any inefficiency will have disappeared.
15
The foreign exchange earnings can be expected to be higher or lower for any of a variety of reasons — change in
terms of trade of some export good or any stochastic shock to the level of foreign exchange reserves. Alternatively, we
can assume that these foreign exchange earnings are drawn from a distribution, since it allows for the possibility, that
even if the foreign exchange reserve situation is quite comfortable, under some contingencies, a large unexpected shock
can precipitate a devaluation. However, we already capture this effect by making z drawn from a continuous distribution.
418
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
while if foreign exchange earnings are expected to be low, then a conservative low expenditure
policy is prudent. We now elaborate.
In particular, in accordance with Krugman (pp. 112, 1999) and Rodrik's (1999) recounting
of currency crises in the nineties, for the remainder of this section we assume that the priors of
the international capital market are skewed in favor of the conservative fiscal policy being as
more likely to be appropriate, such that ω = P(RL) > 1 / 2. Government policymaking consists of
choosing a level of expenditure which be high (xH) or low (xL = 0) and results in a
corresponding high (GH) or low (GL) public good productivity such that output is given by
y = Gjkα. We assume that any government expenditure xjt−1 incurred in period t − 1 that is not
covered by the stochastic foreign exchange earnings Rkt in period t, will be covered by either a
running down of existing foreign exchange reserves or a tax on foreign capital (through
devaluation) in the next period. This implies that the government's two-period budget
constraint is given by xjt−1 = Rkt + τyt where τ is the implicit tax on output as a result of
devaluation. Accordingly, a conservative fiscal policy entails low government expenditure
which is consistent with a balanced budget in each period. In contrast, an aggressive fiscal
policy entails high government expenditure which results in a temporary budget deficit, that
has to be balanced by next period. The appropriateness of either an aggressive or conservative
fiscal policy depends on a government's stochastic budgetary situation (itself a function of the
stochastic foreign exchange earnings). If the foreign exchange earnings are expected to be
‘high’ RH then xH is the ‘appropriate’ policy; if earnings are expected to be ‘low’ RL, then it is
x L , where we assume R H ≥ x H > R L = x L = 0. 16 We ensure this by assuming that
1
1
ðaPðRH jsH ÞGH Þ1−a −ðaPðRL jsH ÞGH Þ1−a zð1 þ rÞxH , where the equality holds when ω = P(RL) =
ϕ, where ϕ → 1. This parametric restriction is a sufficient condition to ensure that on receipt of
signal sH the government's foreign exchange reserves (and hence payoff) increase to a greater
extent when the government chooses xH, than when it chooses xL. Finally, at the time of
making its policy decision, the government receives a private signal sj about the expected
foreign exchange earnings Rj.17
Therefore, the government faces a trade-off. On the one hand, high government expenditure
xH, by boosting public good productivity, has the potential to attract capital inflows and
thereby strengthen fundamentals (i.e. foreign exchange reserves). On the other hand, higher
public expenditure may adversely effect the returns to investment and dampen capital flows, if
it increased the possibility of devaluation. Of course, this is more likely if the government's
budgetary prognosis was bleak and expected foreign exchange earnings failed to materialize.
Similarly, choice of policy xL has the advantage of keeping expenditure low. However, such a
policy is also likely to keep foreign capital inflows restricted to a modest level, with the
currency remaining vulnerable to a bad shock to the foreign exchange situation (i.e. low
realization of Rt).
16
The parallel between the previous framework and the present application become readily apparent when we observe
that in fundamentals are given by θt = (R̂ − xj) + θ̂t−1 + kt where R̂ is a given realized value of foreign exchange reserves.
Once again, there is no devaluation so long as v + θ̂t ≥ z, where the probability of a successful defense is increasing in the
size of capital inflows.
17
This signal can be a privately observed level of foreign exchange reserves which is positively correlated to future
foreign exchange earnings or some private information about the future trajectory of exchange rate earnings.
Alternatively, the government may have private information about a looming political crisis for which it will have to
make a contingent expenditure and draw on reserves.
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
419
Keeping this in mind, consider the payoffs to the government from its equilibrium strategy
σLj⁎. The payoff on receipt of signal sH equals:
⁎ þ wH ½v−z þ h ̂ þ k ⁎ :
̂ þ kLj
wg ðxL ; sH ; r⁎Lj Þ ¼ wLLH ½v−z þ ht−1
t−1
LH
Lj
Here the probability of no devaluation when choice of policy xj on receipt of si when foreign
exchange earnings are Rk, is given by ψjik. The first (second) term on the right hand side is the
government's payoff from defending the exchange rate when foreign exchange earnings turn out
to be RL(RH). Recollect that in our formulation, the government's payoff if the currency peg
collapses is zero. In contrast, the government's payoff from deviating from such an equilibrium
and enacting xH on receipt of sH is given by (with out-of-equilibrium beliefs given by μ):
L
⁎
⁎
̂
̂
wg ðxH ; sH ; r⁎Lj Þ ¼ wH
HH ½v−z þ ht−1 þ kl þ wHH ½v−z þ ht−1 þ kl :
Therefore for a pooling equilibrium to exist, there should exist parameters such that wg(xL,
sH;σLj⁎) ≥ wg(xH,sH;σLj⁎). In the following Proposition we demonstrate that if the priors of the
international capital market are sufficiently skewed in favor of the conservative balanced budget
policy xL, then such a policy is always enacted.
Proposition III. If priors are sufficiently skewed in favor of the pessimistic foreign exchange
earnings forecast RL, then there exist parameters such that there is a pooling perfect Bayesian
equilibrium, where the government enacts the conservative low expenditure policy.
The intuition for the above result is similar to the preceding section. Consider the dilemma
facing the government when ω is sufficiently close to one. It receives information that its foreign
exchange situation is expected to be good. In such a scenario, an activist fiscal policy xH, which
may entail running a temporary budget deficit, will not only increase economic productivity but
may also increase the government's ability to defend the exchange rate peg. However such a
policy, even if socially optimal, also entails some risks for the government. In particular, such a
policy contradicts the priors of the international capital market — which are skewed in favor of
the conservative economic policy as being the appropriate one (since foreign exchange earnings
are expected to be low). Indeed given that such a policy contradicts the common priors, the choice
of policy xH may result in foreign investors (correctly) inferring that the government is likely to be
somewhat less confident of the eventual success of such a policy, and accordingly invest less. The
government is averse to lower foreign capital inflows since by lowering foreign exchange
reserves, it increases the country's vulnerability to an unexpectedly large negative shock (i.e. low
realization of exchange earnings R̂L). In turn, precisely to avoid ‘spooking’ the international
capital market, the government may prefer to conform with the priors of the international capital
market and maintain a conservative policy xL. As a result the economy will be stuck with
relatively modest returns and remain vulnerable to devaluation in the future.
3. Transparency and the IMF
In a world with private information, we observed that governments may have an incentive to
mislead foreign investors. From an operational policy perspective, this immediately gives rise to
the issue of whether we can think of mechanisms which may help mitigate this inefficiency.
420
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
Perhaps the simplest such mechanism would be if we could come up with an institution that
would credibly reveal the government's private information into the public domain.18 However, it
is far from clear whether governments can credibly reveal the private information they receive
under all circumstances.19 Indeed this need for greater transparency and improved data
dissemination was brought home in the aftermath of the Mexican currency crisis, where the
government was believed to have concealed information on the level of foreign reserves. In
response to the Mexican currency crisis, pressure was put on the IMF to create precisely such an
information revealing institution (see Fischer (2002) for a discussion). This resulted in the
introduction in 1996 of the Special Data Dissemination Standard (SDSS) and the General Data
Dissemination Standards (GDSS) by the IMF. Here we discuss the extent to which such measures
are likely to be effective in decreasing (or even eliminating) the likelihood of the ‘confidence
game’ discussed in the previous section.
We now augment the benchmark model to allow for an additional player, namely an
international financial institution (such as the IMF or the World Bank). We assume that this
international institution acquires independent information about the underlying unknown state
of the world in the country in question. The issue of interest is whether this international
institution (the ‘outside’ party) by committing itself to publicly revealing this information, is
likely to reduce or eliminate the government's incentive to engage in the confidence game.
The answer to this depends on the ‘reliability’ (ϕo) of the signal (sjo) received by the
‘outsider’ institution as compared to the common prior (ω) about the country's state of the
world.20 Since the international institution's information is acquired independently of the
country's government, the public announcement of this information by the outside institution
will result in an updated common posterior, ωoj where ωoj = P(SH|sjo) is the new updated
common prior.
o
Suppose an ‘outside’ party such as the IMF received the signal sH
and publicly revealed
this information. Then the government and foreign investors use this information to arrive at
an updated common prior ωoH. To begin with, consider the case where the reliability (ϕ) of
the private signal received by the government is less than this updated common prior i.e.
ωoH > ϕ > 1 / 2. Observe that in this case, quite irrespective of whether the government
receives the private signal sL or sH, the posterior is always unbalanced in favor of the state
being SH, i.e. P(SH|sH,ωoH) > P(SH|sL,ωoH) > 1 / 2. Since the fact that posteriors are balanced in
favor of SH is common knowledge between both investors and the government, the latter
18
In an earlier version of this paper, we argued that one such mechanism may involve delegating decision making
authority to an ideologically contrarian policy maker (as in Cukierman and Tommasi (1998). Since credibility of a policy
is a function of the policy–policymaker pair, we showed that a government that is moderately ideologically predisposed
against the policy favored by international capital markets, is likely to credibly pursue a policy in accordance with its
private signal.
19
One possible mechanism would involve costly signaling — addressed by Rodrik (1989), in the context of trade
liberalization. He argues that credible policy reform may require going inefficiently overboard and much further than it
would have in the absence of a credibility problem.
20
There is no reason why the government's private information set should always coincide with the international
financial institution's. This is perhaps easier to see if we recognize that the set of factors that influence the government's
ability to defend the exchange rate are multi-faceted, and Sk can appropriately be thought of as being a vector with several
components. For instance, the government's ability to successfully defend the exchange rate is a function of a variety of
circumstances — the foreign exchange earnings and the level of reserves, the threat of riots, assassinations (as in the
Colosio assassination in Mexico in 1994) or any other unexpected political imperatives, in addition to information about
the evolution of the economy. Clearly, the IFI's ability to acquire, assimilate and credibly reveal some kinds of
information (e.g. level of foreign reserves), is easier than others (e.g. threat of socio–political unrest).
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
421
will find it optimal to implement policy xH. Clearly in this case, the public announcement of
information by the international institution eliminates the government's incentive to deceive
foreign investors and play the ‘confidence game’ — and thereby improves national income
and welfare.
However, this is not the only possibility. Consider an alternative scenario, where the
private signal received by the government remains more reliable than the updated common
posterior that results from the public announcement that the international institution received
o
sH
, i.e. ϕ > ωoH > 1 / 2. However, if the government's private signal remains more reliable,
then its incentive to deceive foreign investor's through its choice of policy remains (see
Proposition II).
Our above analysis suggests that even if public revelation of information by an international
institution is not sufficient to eliminate the government's incentive to deceive in all instances, it is
unlikely to worsen matters. Unfortunately, this is not quite true. Under some circumstances, it is
possible to go even further and argue that the public release of information by the outside
international institution may even worsen matters.
To see this recall our analysis from Section 2.1. Here the government's incentive to play
the confidence game is highest when the priors of the government and foreign investors are
sufficiently biased in favor of the state being SH. This suggests that if the public release of
information skews beliefs about the underlying state sufficiently, it may result in the
government having an incentive to deceive and thereby lowering expected national income.
To see this let us re-examine Fig. 1. Consider an initial situation where ω = ω̃ < ω̄ , where
ω̃ = ω̄ − ϵ where ϵ is positive and ‘close’ to zero. In this range there is no conflict in the
interests of the foreign investors and the government — as a result the government chooses
the policy that maximizes public good productivity. However, now consider a situation
o
where the international institution acquires information sH
that the state is SH. Since this
information is publicly revealed, suppose that it results in an updated common prior ω̃oH > ω̄ .
This updated common prior is now further skewed. If this skew is sufficiently large, so long
as ϕ > ωoH, it enters the range where the government has an incentive to play the confidence
game. We know this from Proposition II, since this public revelation of information by the
international institution, by skewing the common beliefs sufficiently, results in a
discontinuous change in a government's optimal strategy. Over this range the government
has a clear incentive to play the ‘confidence game’ and will enact policy xH, quite
irrespective of the information that it receives. Expected capital flows (and welfare) drop
discontinuously with the change in optimal strategy.
We summarize the above discussion in the following proposition.
Proposition IV. If ϕ > ϕoj, ∃ϕ > ωo ≥ ω̄ > ω̃, such that a public release of information by the
international financial institution will reduce expected capital inflows and result in the
government's optimal strategy, to switch from a truth-telling separating equilibrium to a pooling
equilibrium.
This result on the welfare impact of greater transparency should be interpreted cautiously.
All that we argue is that an increase in publicly available information has the possibility of
worsening matters. An increase in the public availability of information has two effects. First,
more information, makes it easier for the government to arrive at an accurate assessment of the
underlying state — improving ex–ante welfare. However, more public information has a
second effect — namely, on the government's optimal policy choice. As we saw in the
422
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
preceding discussion — it is this latter effect of an increase in information on the government's
optimal policy choice that is ambiguous. Over some parameter values, an increase in
information may help prevent the government from having an incentive to play the confidence
game — improving welfare. However, over other parameters, an increase in information, may
engender a situation where the government has an incentive to play the confidence game —
which, in equilibrium, has a negative effect on national output. While our above argument is
quite striking, it remains to be examined how applicable this insight is to other contexts and
more general environments. Recent work (in a somewhat different framework) by Morris and
Shin (2002) suggests that it may well be as they argue that “if the agents have access to some
private information, it is not always the case that greater precision of public information is
desirable”.
Therefore, while institutions that encourage greater transparency can be expected to resolve
inefficiencies in governmental policy choices, this may not always be so — indeed, as our
analysis above suggested, in some circumstances it may even worsen matters. As such, we should
not expect greater transparency to be some kind of universal panacea, to tackle inefficiencies in
government decision making. Rather, it should be part of an overall package that encourages
greater accountability in government decision making — by making governmental lengthening
time horizons.
4. Discussion
The factors underlying globalization are multifaceted. Globalization may be a function of
conscious policy choice, as when governments lower barriers to capital flows, or it may be a
result of technological and institutional changes — as in the case of greater information
availability and transparency. Clearly this globalization of capital flows and information
availability has yielded many economic benefits. Indeed, the threat of capital flight itself has
the potential to be an effective disciplining device (Obstfeld, 1998). However, our analysis
suggests that these forces of globalization, while providing opportunities, also bring with it
some risks. For instance, it is only if the globalization of capital flows proceeds far enough, and
governments have the possibility of attracting foreign capital, will they have an incentive to
play the confidence game. The globalization of information and greater transparency, increases
investment and often reduces the government's incentive to make inefficient policy choices.
However, once again, they may be cases where this globalization of information may worsen
matters.
This suggests that at least to some extent, the beneficial impact of the globalization of capital
flows (say due to efficiency in resource allocation) may be dampened, due to the greater incentive
to play the confidence game. Our results should throw a cautionary note to the wave of economic
reforms that have been enacted worldwide.21 Strikingly enough, our analysis suggests that a
change in the degree of ‘consensus’ (increasing skew in priors), coupled with the prospect of
attracting capital from international capital markets may have been responsible for this enthusiasm
for economic reform in at least a few instances. Indeed Krugman (1998), Radelet and Sachs (1998)
and Stiglitz (2002) have all argued that Washington and the IMF have made the mistake of
21
As Rodrik (1996) puts it ‘while the reforms were at least in part inspired by the East Asian experience, they took place
much more quickly and, in many areas, are going considerably beyond those undertaken in East Asia. This raises the
question whether the economic reformers have internalized the correct lessons from the East Asian experience.”
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
423
advocating policies that were directed more at appeasing international markets and winning their
‘confidence’ rather than addressing economic fundamentals. Not surprisingly, it is argued, that
these policies do not seem to have been too successful. Our analysis suggests a reason why
governments may feel ‘disciplined’ into enacting such inefficient policies.
It is important to observe that our analysis is of much greater relevance to the case of
developing and transition economies. This is because greater political instability and the
associated shorter political horizons is likely to give governments in these countries more of an
incentive to deceive. The establishment of political parties is likely to result in the time horizons
of governments increasing — such that the standard reputational concerns that mitigate these
inefficiencies become operational.
Acknowledgements
This paper has benefited greatly from the comments of two referees, the editor Bob Staiger and
conversations with Russ Cooper, Avinash Dixit, Parikshit Ghosh, Francisco Gonzalez, Debbie
Minehart, Dilip Mookherjee, Dani Rodrik, Zvika Neeman, Alwyn Young and especially Jonathan
Eaton. This paper is adapted from the second chapter of Mukand (1998).
Appendix A
Note: In what follows we make use of the following: (i) conditional on receiving sj, expected
productivity of the public good differs across the two states, because either (a) ω > 1 / 2, (b)
GH > GL. (ii) In any equilibrium, higher capital flows, implies higher θ̂ t. (iii) z is drawn from a
continuous distribution on the real line with cdf F(z), and is realized after the government's policy
choice. This implies that since a devaluation of an exchange rate takes place only if v + θ̂ t < ẑ t, we
have ψij bounded away from zero or one. Higher θ̂ t ⇒ lower probability of devaluation ψij and
higher expected payoff to the government. This has an important implication for what follows —
namely, an increase in capital inflows, will ceteris paribus, result in a higher probability of the
government carrying out no devaluation.
Proposition I. The argument is straightforward. So long as ϕ > ω, we know from Assumption 1
that P(Sj|sj) > P(Sj|sk) where j ≠ k. Therefore, if government does not follow signal, expected
productivity of the public good is given by P(Sj|sk)Gj and investor will optimally choose kij⁎ which
is (from Observation 1) is strictly lower than the amount of capital invested if the government
chose to follow its signal. Observe that ceteris paribus, fundamentals θt, are increasing in the total
volume of capital inflows. This immediately implies that both the probability of no devaluation, as
well as the payoff contingent on no devaluation is increasing in capital inflows. Therefore, when
the signal is public, choosing a policy that contradicts the signal, lowers expected productivity of
the public good and adversely affects capital flows and fundamentals. Therefore, in an equilibrium,
an optimizing government will always prefer to pursue a policy in accord with its public signal.
Proposition IIA and IIB. We begin by examining the parameters under which there exists a
truth-telling separating equilibrium, where a government chooses xj on receipt of sj, i.e. the
government's equilibrium strategy is σjj⁎. In such an equilibrium, the government's payoff from
pursuing policy xL after receipt of sL is given by,
⁎ Þ þ w0 ½v− z¯ þ h ̂
̂ þ PðSL jsL ÞGLg ðkLL
wg ðxL ; sL ; r⁎jj Þ ¼ wLLL ½v− z¯ þ ht−1
t−1
LL
⁎ Þ:
þ ð1−PðSL jsL ÞÞG0 gðkLL
424
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
In contrast the government's incentive from deviating from the proposed truth-telling
equilibrium and choosing policy xH on receipt of sL is,
0
⁎
̂
̂
wg ðxH ; sL ; r⁎jj Þ ¼ wH
HL ½v− z¯ þ ht−1 þ PðSH jsL ÞGH gðkHH Þ þ wLL ½v− z¯ þ ht−1
⁎
þ PðSL jsL ÞG0 gðkHH Þ:
Therefore, a truth-telling separating equilibrium exists, when wg(xH,sL,σjj⁎) ≤ wg(xL,sL,σjj⁎).
Rearranging the two preceding equations, we obtain the following inequality:
0
L
0
⁎
⁎
½v− z¯ þ ht−1 ðwH
HL ðrjj Þ þ wHL ðrjj Þ−wLL ð:Þ−wLL ð:ÞÞ
0
L
0
⁎
⁎
⁎
⁎
þ PðSH jsL Þ ½wH
HL GH gðkHH Þ−wLL G0 gðkLL ÞVPðSL jsL Þ½wLL GL gðkLL Þ−wHL G0 gðkHH Þ:
ð3Þ
It is easy to see that there exist parameters such that this inequality is true. First, we examine the
conformity bias. We ensure this by assuming that GH = GL and there is no possibility of the ‘good
⁎ ≈ kLL
⁎ . Since we can choose v − z̄ +
news’ bias. If ϕ → 1 and ω → 1 / 2, then P(SL|sL) → 1 and kHH
θ̂ t−1 to be ‘small’, it is easy to check that this implies that there exists a large set of parameters
for which LHS < RHS. Also observe that the posterior P(SL|sL) is decreasing in ω, i.e.
PðSL jsL Þ
<0.
dx
Further, observe that if ω is close enough to one, i.e. ϕ > ω → 1, then P(SH|sH) → 1 > P
⁎ > kLL
⁎ . Since θ̂ t is ceteris paribus,
(SL|sL) → 1 / 2. This immediately implies that kHH
H
L
increasing in capital flows, this asymmetry in capital inflows, in turn implies that ψHL
> ψLL
.
Ipso facto, this immediately suggests that there exist parameters such that LHS > RHS.
Furthermore, given that both the LHS (and RHS) are continuous and monotonic increasing
(decreasing), we have a cut-off ω̂ such that for all ω < ω̂ there exists a truth-telling separating
equilibrium.
Similarly, consider the condition for a pooling equilibrium at xH for the case where GH = GL.
For a pooling equilibrium to exist, there must exist parameters such that wg(xH,sL,σHj⁎) ≥ wg(xL,
sL,σHj⁎,μ), where on observing xH, the foreign investor's out-of-equilibrium beliefs about the
private signal received by the government, are given by μ. This gives rise to:
0
L
0
⁎
⁎
⁎
⁎
½v−z̄ þ ht−1 ðwH
HL ðrHj Þ þ wHL ðrHj Þ−wLL ðrl Þ−wLL ðrl ÞÞ
H
0
⁎
⁎
⁎ Þ−w0 G gðk ⁎ Þ:
þ PðSH jsL Þ ½wHL GH gðkHj Þ−wLL G0 gðkLl ÞzPðSL jsL Þ½wLLL GL gðkLl
HL 0
Hj
ð4Þ
Once again, observe that if ϕ → 1 and ω → 1 / 2, then P(SL|sL) → 1 and P(SH|sL) → 0. This
immediately implies that if we pick a z such that (v − z̃ + θt−1) is ‘small’ and negative, we will
obtain LHS < RHS. Now once again simple comparative statics on ω show that the inequality
reverses itself for ϕ > ω → 1. Here, we have P(SH|sH) → 1 > P(SL|sL) → 1 / 2. Further observe that
⁎ > kLμ
⁎ , even if the investor's
since ω → 1 and the consequent unbalanced posteriors, we have kHj
out-of-equilibrium belief on observing xL is that the government received the private signal sL.
⁎ ) > g(KLμ
⁎ ), which given that θt is an increasing function, implies that
This implies that g(kHj
H
L
ψHL > ψLL. Since in addition P(SH|sL) ≈ P(SL|sL) → 1 / 2, there exist parameters such that
LHS > RHS. Given continuity of the underlying functions, there exists a pooling equilibria for
all ω ≥ ω̄ .
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
425
Finally, we want to evaluate whether ω̄ is greater, less than or identical to ω̂. For convenience we
rewrite Eq. (3) above to obtain a condition for a deviation from a truth-telling separating equilibrium:
0
L
0
⁎
⁎
⁎
⁎
A1 þ PðSH jsL Þ½wH
HL GH gðkHH Þ−wLL G0 gðkLL Þ−PðSL jsL Þ½wLL GL gðkLL Þ−wHL G0 gðkHH Þz0:
ð5Þ
H
0
L
0
Here A1(σjj⁎) = [v −¯z + θt−1](ψHL
+ ψHL
− ψLL
− ψLL
). Similarly, for a pooling equilibria at xH, we
H
0
L
0
⁎ ) = [v −¯z + θt−1](ψHL
rewrite Eq. (4) above, where A2(σHj
+ ψHL
− ψLL
− ψLL
) as,
0
L
0
⁎
⁎
⁎
⁎
A2 þ PðSH jsL Þ½wH
HL GH gðkHj Þ−wLL G0 gðkLl Þ−PðSL jsL Þ½wLL GL gðkLl Þ−wHL G0 gðkHj Þz0
ð6Þ
Now we are interested in comparing Eq. (5) with Eq. (6). First, we can always choose v,¯z so as
to ensure that A1 and A2 are ‘negligible’. Now examine both these inequalities as we increase ω
⁎ > kHj
⁎ , the first inequality that will be violated is the truth telling
from 1 / 2. Observe that since kHH
condition (5). Given that we can always choose G0, small enough, this immediately suggests that
ω̂> ω̄ .
Similarly, consider the case of the good news bias. Here we assume ω = 1 / 2 and GH > GL so as
to ensure that there is no skew in the priors. If ω = 1 / 2, then we can always pick a ϕ large enough
to ensure that LHS ≤ RHS. Now having pinned ϕ down we are still free to choose GH. Using an
argument similar to the one used earlier which utilizes the continuity and monotonicity of the LHS
and RHS of Eq. (4), we can show that as we increase GH, there exists a pooling equilibria where
the government enacts xH, for all GH ≥ Ḡ H.
Proposition III. We now sketch the Proof of Proposition III, which in structure is similar to the
preceding analysis.
For a pooling equilibrium, we need to demonstrate whether there exist parameters such that wg
(xL,sH,σLj⁎) ≥ wg(xH,sH,σLj⁎). From the discussion in the text, we have,
⁎ þ wH ½v− z¯ þ h ̂ þ k ⁎ :
̂ þ kLj
wg ðxL ; sH ; r⁎Lj Þ ¼ wLLH ½v− z¯ þ ht−1
t−1
LH
Lj
ð7Þ
The government's payoff from deviating from such an equilibrium and enacting xH on receipt
of sH is given by (with out-of-equilibrium beliefs given by μ):
L
⁎
⁎
̂
̂
wg ðxH ; sH ; r⁎Lj Þ ¼ wH
HH ½v−z þ ht−1 þ kl þ wHH ½v− z¯ þ ht−1 þ kl :
ð8Þ
Rearranging the preceding equations, we obtain,
⁎ Þ½wL þ wH −ðv− z¯ þ h ̂ þ k ⁎ Þ½wH þ wL z0:
̂ þ kLj
ðv− z¯ þ ht−1
t−1
LH
LH
HH
HH
l
In the two preceding equations, since the production function is Cobb–Douglas, kLj⁎ =
[αωGL]1/(1−α) and kμ⁎ = [αμGH]1/(1−α), where μ are the out-of-equilibrium beliefs of the
investor — on seeing a deviation to xH, we assume that the investor believes that the
government received the sH signal. Now, recollect that the government will not devalue so
L
H
⁎)[ψLH
long as v + kt + θ̂ t−1 ≥¯z . Now consider Z1 − Z2 ≥ 0, where Z1 = (v − z + θ̂ t−1 + kLj
+ ψHL
] and
H
L
Z2 = (v − z + θ̂ t−1 + kμ⁎ )[ψHH + ψHH]. We compare Z1 with Z2 by first comparing pairwise the
L
two terms in the square brackets. Here ψLH
= P(v + θt ≥¯z ) is increasing in both v and θt. Now
H
⁎
θt = RL − xL + θ̂ t−1 + kLj. Similarly, ψHH = P(v + θt ≥ z) where θt = RH − xH + θ̂ t−1 + kμ⁎. Observe that
426
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
L
H
in expected terms, this implies that ψLH
≥ ψHH
iff RL − xL + kLj⁎ ≥ RH − xH + kμ⁎, which equals
H
kLj⁎ > kμ⁎ since xj = Rj for j ∈ {L,H}. Likewise, ψHL
= P(v + θt ≥¯z ) where θt = RH − xL + θ̂ t−1 + kLj⁎
H
H
H
and ψHH
= P(v + θt ≥¯z ) where θt = RH − xH + θ̂ t−1 + v + kμ⁎. Once again observe that ψLH
≥ ψHH
if
kLj⁎ > kμ⁎. Therefore, not only is the size of Z1 − Z2 a function of the gap between kLj⁎ and kμ⁎,
but further if kLj⁎ > kμ⁎, then we have a sufficient condition to ensure that Z1 > Z2.
Moreover, notice that all the ψijk above, are increasing in the size of capital inflows.
Further, notice that since dZ1 / dω > 0, and dZ2 / dω < 0, the left hand side above is increasing with
dðZ1 −Z2 Þ
> 0. Now we need τ to be large enough to ensure that if devaluation
ω and we have
dx
occurs, then there is a sufficient adverse impact on capital flows. Keeping in mind that P(RH|sH) →
1 / 2 for ω → ϕ, when ϕ is close to one, we ensure this by checking to see whether the following
condition is satisfied:h 1=2½alG
̂ iH þ 1=2½ð1−sÞaPðRH jsH ÞGH < ½axGL . This holds so long as
xH > GH ðaPðRH jsH ÞGH Þ 1− 2 ̂xG . Observe that if ω → ϕ and ϕ → 1, we have (ω / μ̂) → 1. Further
lG
since GH > GL, this implies that the above inequality should be satisfied for a large number of
parameter values. We further can check to see that upper bound on xH imposed by assumption is
not violated for GH > GL. Hence, as ω → 1, we have kLj⁎ > kμ⁎. This directly implies that there exists a
non-empty set of parameters such that our pooling equilibrium holds, where the government
prefers xL.
1
1−a
1
1−a
1−a
1
1−a
1
1−a
L
H
References
Angeletos, George-Marios, Werning, I., 2005. Crises and Prices: Information Aggregation, Multiplicity and Volatility,
MIT Working Paper.
Bartolini, Leonardo, Drazen, Allan, 1997. Capital-account liberalization as a signal. American Economic Review 87 (1),
138–154.
Bhagwati, Jagdish, 1998. The capital myth — the difference between trade in widgets and dollars. Foreign Affairs 77 (3),
7–12.
Brandenburger, Adam, Polak, Ben, 1996. When managers cover their posteriors — making decisions the market wants to
see. Rand Journal of Economics 27 (3), 523–541.
Cai, Hongbin, Treisman, Daniel, 2005. Does competition for capital discipline governments? Decentralization,
globalization, and public policy. American Economic Review 95 (4) (June).
Camdessus, Michael, 1998. The IMF and its Programs in Asia, remarks delivered at the Council of Foreign Relations, New
York, Feb. 6, 1998.
Cukierman, Alex, Tommasi, M., 1998. When does it take a Nixon to go to China? American Economic Review 88 (1).
de Kock, G., Grilli, Vittoro, 1993. Endogenous exchange rate regime switches. Economic Journal 103, 347–358.
Dornbusch, Rudiger, 2000. Stiglitz versus the IMF: Another View, mimeo. MIT.
Drazen, Allan, 1997. Policy signaling in an open economy: a reexamination. In: Wolf (Ed.), Holger. Contemporary
Economic Development Reviewed, vol. 5. Macmillan, London.
Feldstein, Martin, 2000. Aspects of Global Economic Integration, NBER Working Paper 7899.
Fischer, Stanley, 2002. Financial Crises and Reform of the International Financial System, NBER Working Paper 9297.
Krugman, Paul, 1998. The Confidence GameThe New Republic. October 5 1998.
Krugman, Paul, 1999. The Return of Depression Economics. W.W. Norton and Co.
Krugman, Paul, Venables, A., 1995. Globalization and the inequality of nations. Quarterly Journal of Economics 110,
857–880.
Majumdar, Sumon, Mukand, Sharun, 2004. Policy gambles. American Economic Review 94 (2).
Morris, Stephen, Shin, Hyun Song, 1998. Unique equilibrium in a model of self-fulfilling currency attacks. American
Economic Review 88 (3).
Morris, Stephen, Shin, Hyun, 2002. Social value of public information. American Economic Review 92, 1521–1534.
Mukand, Sharun Williams, 1998. Domestic Imperatives and the International Arena. PhD Dissertation, Boston University.
Mukand, Sharun, Rodrik, Dani, 2005. In search of the holy grail: policy convergence, experimentation and economic
performance. American Economic Review 95 (1) (March).
Obstfeld, Maurice, 1994. The Logic of Currency Crises, NBER Working Paper 4640.
Obstfeld, Maurice, 1996. Models of Currency Crises with Self-fulfilling Features, European Economic Review, April.
S.W. Mukand / Journal of International Economics 70 (2006) 406–427
427
Obstfeld, Maurice, 1998. Global capital markets: benefactor or menace. Journal of Economic Perspectives 12 (4), 9–30.
Ozkan, G., Sutherland, A., 1995. Policy measures to avoid a currency crises. Economic Journal 105, 510–519.
Prendergast, Canice, Stole, Lars, 1996. Impetuous youngsters and jaded old-timers: acquiring a reputation for learning.
Journal of Political Economy 104, 1105–1134.
Radelet, Steve, Sachs, Jeffrey, 1998. The East Asian Financial Crisis: Diagnosis, Remedies and Prospects, mimeo. HIID.
Rodrik, Dani, 1989. Promises, promises: credible reform via signaling. Economic Journal 99.
Rodrik, Dani, 1996. Journal of Economic Literature 34, 9–41.
Rodrik, Dani, 1998. Who Needs Capital Account Convertibility? Working Paper, Kennedy School of Government,
Harvard University.
Rodrik, Dani, 1999. Governing the Global Economy: Does One Architectural Style Fit All? Brookings Trade Forum.
Brookings, Washington D.C.
Rogoff, Kenneth, 1999. International Institutions for Reducing Global Financial Instability, NBER Working Paper 7265.
Sachs, J., Tornell, A., Velasco, Andres, 1996. The collapse of the mexican peso: what have we learned? Economic Policy
22.
Stiglitz, Joseph, 2002. Globalization and its Discontents. W.W. Norton, Washington DC.
Summers, Lawrence, 2000. International financial crises: causes, prevention and cures. American Economic Review 90
(2).
Wade, Robert, Veneroso, Frank, 1998. The Asian Crisis: The Hight Debt Model vs. The Wall Street-Treasury-IMF
Complex, mimeo. The Russell Sage Foundation.